Daemon process hammering disk

[Adamski]

Hi all,
I've just put a fresh version of Ubuntu 10 Server (was previously running CentOS) on my server PC this week and installed sabnzbdplus via apt-get. The process was running just fine for the first few NZBs that I downloaded... However, now it seems that whenever it tries to download something new, the disk busy % goes through the roof and the server becomes VERY slow... The only way to get the server working again is to shutdown sabnzbd... I never had any problems on CentOS, so I tried copying the original config file that I was using on there, however this doesn't see to make any difference.

I've seen a couple of threads on 100% CPU usage, but nothing relating to disk issues   Any ideas what I could do?

Code: Select all

root@monkey:/etc/default# dpkg --list | grep sab
ii  sabnzbdplus                     0.5.5-0ubuntu1~jcfp1~maverick       web-based binary newsgrabber with nzb support
ii  sabnzbdplus-theme-classic       0.5.5-0ubuntu1~jcfp1~maverick       classic interface templates for the SABnzbd+ binary newsgrabber
ii  sabnzbdplus-theme-mobile        0.5.5-0ubuntu1~jcfp1~maverick       mobile interface templates for the SABnzbd+ binary newsgrabber
ii  sabnzbdplus-theme-plush         0.5.5-0ubuntu1~jcfp1~maverick       plush interface templates for the SABnzbd+ binary newsgrabber
ii  sabnzbdplus-theme-smpl          0.5.5-0ubuntu1~jcfp1~maverick       smpl interface templates for the SABnzbd+ binary newsgrabber

sabnzbd.ini can be posted if needed

Looking in the log file. This is the last thing it logged before I shut it down:

Code: Select all

2010-11-18 18:37:28,635::INFO::[downloader:654] Connecting [email protected]:443 finished
2010-11-18 18:37:29,005::INFO::[downloader:654] Connecting [email protected]:443 finished
2010-11-18 18:37:29,916::DEBUG::[decoder:92] Decoding <Article: [email protected], bytes=397350, partnum=1, art_id=None>
2010-11-18 18:37:29,921::INFO::[downloader:465] [email protected]:443: Initiating connection
2010-11-18 18:37:31,080::INFO::[downloader:465] [email protected]:443: Initiating connection

Thanks

Adam

[shypike]

Hard to tell from so little info.
What you can try is to use a decent article cache (say 80M)
and set logging to the minimum.
I can think of no real reason for it to start hammering the disk,
except for par2 repairs and unrars.

[Adamski]

Thanks. I did have the article cache set to 100M, but didn't seem to resolve the problem.  Log level is also set to 2. The problem seems to be when actually downloading articles before it even gets to the RAR / PAR2 stage

[jcfp]

In addition to that, the excessive number of connections also doesn't do any good, unless your internet connection is a private OC-12 line.

[Adamski]

In addition to that, the excessive number of connections also doesn't do any good, unless your internet connection is a private OC-12 line.

Excessive? You mean 35?

I have gone back to using the SAB directory that I had on my CentOS install now.. It all works, however I get the exact problem when downloading again. I have always had 50 connections set up for my usenet provider and not had a problem before.

[shypike]

Weird.
In the last 4 versions of SABnzbd there have been 0 changes
in the download code.
Just minor logic errors in other parts.
I'm afraid I cannot explain this.
I can start blaming the installed modules (Python, Cheetah, PyOpenSSL etc.)
but then more people would be reporting similar problems.
I still think you should reduce the logging level to the minimum.